Research And Implement On The Distributed Constructing Of Concept Lattice

Concept lattice is a perfect mathematical model which describes knowledge with the relation between the intensions and extensions of concepts, and the relation between the generalization and specialization of concepts. It's good to show the data and knowledge and describe many questions of knowledge discovery, such as classification, association rules, sequential patterns and clustering. As the size of data growing fast, the distributed constructing of concept lattice becomes an important research subject. Therefore, it becomes the major subject in this dissertation. The contributions of the dissertation are as follows:1) The Sub-Top Concept Lattice (STCL) is proposed, which is a distributed constructing of concept lattice based on the Sub-Top Concept. Firstly, generate all Sub-Top concepts, divide the database into Sub-Top contexts and construct the Coping of lattice on the Master Node; then, construct Sub-Top lattices on the Slave Nodes; finally, merge all Sub-Top lattices and the Coping of lattice to a complete lattice on the Master.2) The Intension Reduction STCL (IRSTCL) is proposed, which is an extension of STCL. STCL divides the context only by the Object set, but the extension of STCL divides the context by both Object and Character sets. The analysis shows that the time and space performance of IRSTCL are better than STCL.3) A knowledge discovery archetypal system named KDDCL is designed. The system can deal with many tasks of Data Mining, which integrate the concept lattice model and some extended models of concept lattice. Now, the core of the system is completed, including the module of data preparation, concept lattice construction and graphical display of concept lattice.